Jeffrey R. Irwin Victoria M. Scholl Mahesh Shrestha Travis J. Kropuenske Josip D. Adams Matthew A Burgess Lance R. Brady Aparajithan Sampath 20251118 remote-sensing image Jeffrey R. Irwin ORCiD 0000-0001-5828-0787 Victoria M. Scholl 0000-0002-2085-1449 Mahesh Shrestha ORCiD 0000-0002-8368-6399 Travis J. Kropuenske ORCID 0000-0002-3269-4225 Josip D. Adams 0000-0001-8470-4141 Matthew A. Burgess 0000-0003-3487-4972 Lance R. Brady 0009-0000-4233-3856 Aparajithan Sampath 0000-0002-6922-4913 https://doi.org/10.5066/P14FFHW5 Mahesh Shrestha Victoria Scholl Aparajithan Sampath Jeffrey Irwin Travis Kropuenske Josip Adams Matthew Burgess Lance Brady 20251117 publication Remote Sensing vol. 17, issue 22 n/a MDPI AG ppg. 3738 https://doi.org/10.3390/rs17223738 U.S. Geological Survey (USGS) scientists and engineers from the Earth Resources Observation and Science Center (EROS) Calibration/Validation Center of Excellence (ECCOE) and the National Uncrewed Systems Office (NUSO) conducted field data collection efforts July 15 – 18, 2023 at the USGS EROS facility in Minnehaha County, South Dakota. The data collected included multispectral (MS) and hyperspectral (HS) imagery from two commonly used Uncrewed Aircraft Systems (UAS) mounted cameras, as well as field spectrometer measurements collected from the ground. The work was initiated to assess the radiometric quality of the imagery collected by the MS and HS cameras when the cameras were calibrated using their respective manufactures’ recommended procedures and to see if the quality could be improved by implementing the Empirical Line Method (ELM). Additionally, several different calibration targets were deployed to determine the impacts of target size, material, reflectance intensity, and quantity on the ELM for UAS data. Only the data from the noon collection on July 18 is provided in this data release as the other collections were impacted by wildfire smoke, clouds, aerosol, and/or gimbal issues. See the associated journal publication for further details. This field data collection exercise was run with a dual focus: to study the calibration and validation capabilities of sensors mounted on UAS, and to evaluate how these UAS sensors can be used to validate surface reflectance products derived from satellite data. Numerous radiometric sensors are compact and lightweight enough to be deployed on UAS platforms. All radiometric sensors, whether high-end satellite or aircraft sensors, initially collect data in a format known as Digital Numbers (DNs). While satellite spectral data is often distributed as reflectance—indicating the percentage of light that reflects off a surface—UAS radiometric data can also be converted to reflectance using calibration panels with known reflectance values. A key distinction between high-end satellite sensors and those used on UAS is that satellite sensors are typically calibrated in laboratory settings and often include onboard calibration systems. Given the impracticalities of lab calibration for every UAS sensor, a more effective strategy is to calibrate the imagery captured by these sensors during field operations. To support this, ECCOE, in collaboration with NUSO, has developed the Guidelines for Calibration of Uncrewed Aircraft Systems Imagery (https://doi.org/10.3133/ofr20231033), this exercise expands upon that foundational work. The data obtained through ScienceBase at https://www.sciencebase.gov/catalog/item/682b85aad4be024992ebd701 are considered the "best available" data from the USGS. For questions on distribution, please refer to the Distribution Section, Contact Information. For processing, please refer to the Data Quality Section, Processing Step, Contact Information. 20230718 ground condition Complete None planned -96.63918 -96.60416 43.75213 43.72385 ISO 19115 Topic Category imageryBaseMapsEarthCover None Field Spectrometer Surface Reflectance Surface Reflectance Validation Imagery Calibration Multispectral Imagery Hyperspectral Imagery Uncrewed Aircraft System UAS Calibration Targets Empirical Line Method USGS Metadata Identifier USGS:682b85aad4be024992ebd701 None South Dakota Minnehaha County USGS EROS Center Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data for other purposes, nor on all computer systems, nor shall the act of distribution constitute any such warranty. Although these data have been processed successfully on a computer system at the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data for other purposes, nor on all computer systems, nor shall the act of distribution constitute any such warranty. The USGS or the U.S. Government shall not be held liable for improper or incorrect use of the data described and/or contained herein. Jeffrey R. Irwin U.S. Geological Survey Earth Resources Observation and Science Center Geographer mailing address

47914 252nd Street

Sioux Falls SD 57198 United States 605-594-2731 605-594-6529 jrirwin@usgs.gov Environment as of Metadata Creation: Microsoft Windows 11 Enterprise Version 10.0.22631 Build 22631 The Analytical Spectral Device FieldSpec 4 is calibrated by Analytical Spectral Devices, Inc. The device is identified as having a radiometric calibration accuracy of +/- 5% from 400 - 900 nm, increasing to +/- 8% at 2200 nm. A field laptop is used with the Fieldspec 4 for recording of radiometric data and integration of Coordinated Universal Time (UTC) from the laptop. The laptop establishment of UTC varies, with an estimated accuracy at +/- 30 seconds (clock drift between monthly synchronizing events). The Spectralon Panels used within the field collection have average uncertainties of 0.0064 (250-600nm), 0.0042 (601-1500nm), 0.0107 (1501-2200nm), and 0.0413 (2201-2500nm); (uncertainty listed is average of reflectance range from 0.00 to 0.99). The GlobalSat BU-353S4 USB GPS Receiver has a horizontal accuracy of about 12 m. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. The dataset is considered complete for the information presented, as described in the abstract. Users are advised to read all metadata of the record carefully for additional details. The dataset is considered complete for the information presented, as described in the abstract. Users are advised to read all metadata of the record carefully for additional details. ECCOE Single Spectrometer Data Processing Steps. Field spectral data were collected in Digital Numbers (DN) using the RS3 version 6.4.0 software from Malvern Panalytical and stored to a laptop connected to the spectrometer. During the field data collection, panels of known reflectance were measured with the spectrometer periodically. These panels were calibrated at the University of Arizona, College of Optical Sciences, Bidirectional Reflectance Distribution Function (BRDF) facility. Spectral measurements of the natural target were taken as transects between the measurements of the reference panels. Natural target spectral data was tagged for location with an autonomous grade Global Positioning System (GPS) device connected to the laptop during data collection. DNs were converted to radiance values with Malvern Panalytical’s ViewSpec Pro version 6.2.0 software. The reference panel data were interpolated between each panel measurement, creating an estimated reference panel measurement for each spectrometer measurement of the natural target. Surface reflectance values for each spectrum were calculated by taking the ratio of the radiance of the natural target and the radiance of the interpolated reference measurement of the target and multiplying the ratio by the calibration coefficient. 20250106 ECCOE Dual Spectrometer Processing Steps. The dual spectrometer approach utilizes two field spectrometers, a mobile unit (MU) and a fixed base unit (FBU). Field spectral data were collected in DNs using the RS3 version 6.4.0 software from Malvern Panalytical and stored to a laptop connected to the spectrometer for both the MU and FBU. During the field data collection, panels of known reflectance were measured with the spectrometer periodically with the MU while the FBU continuously measured a panel of known reflectance. These panels were calibrated at the University of Arizona, College of Optical Sciences, BRDF facility. Spectral measurements of the natural target were taken as transects between the measurements of the reference panels. Spectral measurements of the natural target were taken as transects between the measurements of the reference panels with the MU. Natural target spectral data was tagged for location with an autonomous grade Global Positioning System (GPS) device connected to the MU laptop during data collection. DNs were converted to radiance values with Malvern Panalytical’s ViewSpec Pro version 6.2.0 software for both the MU and FBU. Data collected with the MU included measurements of the reference panel that the FBU was monitoring, as well as measurements of the MU reflectance panel and the natural target. The coincident measurements taken by the MU and FBU of the FBU reference panel were used to establish the initial reflectance calibration gain of both the MU and FBU. This was calculated by taking the ratio of the FBU panel reflectance and radiance measured by both the MU and FBU. The MU reflectance calibration gain was updated continuously for the measurements of the natural target using the FBU measurements. First the radiance collected by the MU was adjusted by multiplying the radiance by the ratio of the FBU radiance at a given time and the FBU radiance at the first coincident measurement of the FBU reference panel. The reflectance calibration gain of the MU was then calculated/updated by taking the ratio of the FBU reflectance and the adjusted MU radiance. The surface reflectance of the natural target was then determined by multiplying the radiance of the natural target by the updated reflectance calibration gain. 20250106 NUSO Multispectral Imagery Processing Steps Low-altitude UAS flights were conducted by the USGS National Uncrewed Systems Office (NUSO) at the Earth Resources Observation and Science (EROS) Center in Sioux Falls, SD. A DJI Matrice 600 Pro UAS with approved government edition firmware carrying a MicaSense RedEdge-MX Dual sensor was flown at an altitude of 62 meters (200 feet) above ground level to capture multispectral images with automatic exposure settings. The Dual configuration includes a downwelling light sensor (DLS 2) secured to the top of the UAS, which captured data about the incident illumination during flight. The DLS2 values are stored in metadata for each image. A MicaSense-provided calibration panel (serial number RP06-2210452-OB) was imaged before and after each UAS flight as recommended by the manufacturer. The first UAS transect was flown East to West over a row of reference panels. The aircraft then transitioned to flying North-South transects over the vegetated field. Survey control was established using Propeller AeroPoint temporary ground control points (GCPs) distributed throughout the survey area. GCPs were post-processed with corrections from a concurrently operating Trimble R8s GNSS base station. Multispectral images were processed in Agisoft Metashape Professional (version 2.0.4) photogrammetry software using the general workflow outlined in Over and others (2021). A separate project was created for each flight. Photos captured at flight altitude were imported into the software, aligned with high accuracy, and an initial optimization was performed. In the reference settings, camera accuracy was set to 2 degrees, marker (GCP) accuracy was set to 0.01/0.02 m based on calculations using AeroPoint data, image marker accuracy was set to 0.5 pixels, and the project coordinate system was converted from the default geographic system to a projected coordinate system, NAD83(2011) / UTM zone 14N (EPSG 6343). GCPs were imported as markers and their locations were manually refined to match the center of each target location within the images. The point cloud (also known as tie points, resulting from the photo alignment and optimization) was edited using an iterative error-reduction procedure to filter out points with high errors. This error-reduction was done in several iterations of a process called "Gradual Selection," first to reduce reconstruction uncertainty ("RU", to a unitless value of 10), then projection accuracy ("PA", to a weighted value of 3), and finally reprojection error ("RE") of 0.3 for the tie points. A digital surface model (DSM) raster was generated using the remaining tie points. Using photos containing the MicaSense calibration panel, images were then calibrated to reflectance following manufacturer guidelines. Reflectance calibration was run both with and without the DLS2 incorporated, as it was determined to have variable impacts on the output values. An orthomosaic was then generated using the resulting 16-bit reflectance images and the DSM surface, default Mosaic belnding mode, and hole filling enabled. Multispectral bands 1 through 10 of the orthomosaic were divided by a scale factor of 32768 using the Raster Calculator tool to yield a 10-band reflectance raster product with values between 0-1, where 1 indicates 100% reflectance. The 10-band reflectance orthomosaic rasters were exported at native resolutions (approximately 4 cm) with projected coordinate system NAD83(2011)/UTM Zone 14N, in .tif format. In ENVI software version 5.6, a header file (.hdr) was generated to accompany each orthomosaic to include respective band names and center wavelength information for each of the 10 multispectral spectral bands. References: Over, J.S.R., Ritchie, A.C., Kranenburg, C.J., Brown, J.A., Buscombe, D.D., Noble, T., Sherwood, C.R., Warrick, J.A. and Wernette, P.A., 2021. Processing coastal imagery with Agisoft Metashape Professional Edition, version 1.6—Structure from motion workflow documentation (No. 2021-1039). US Geological Survey. 20250310 NUSO Hyperspectral Imagery Processing Steps Headwall Nano-Hyperspec hyperspectral UAS flights were flown at an altitude of 62m (200ft) above ground level. The sensor captured 274 spectral bands spanning the visible and near infrared wavelengths, 398 to 1002 nm and had a 12 mm focal length lens. A dark reference image was collected prior to each flight while the sensor lens cap was on. Sensor exposure was set based on ambient illumination levels prior to each flight using a white piece of paper as a reference to avoid saturation of bright materials. The sensor was flown with an integrated Applanix APX-15 GNSS inertial measurement unit to record the sensor's location and orientation during each flight. The first UAS transect was flown East to West over a row of reference panels. The aircraft then transitioned to flying North-South transects over the vegetated field. The hyperspectral images were post-processed using the sensor manufacturer's proprietary software and following their recommended workflow as summarized here. In Headwall SpectralView (Hyperspec v3.2.0 - Airborne) software, raw hyperspectral images were radiometrically converted to units of radiance (mW/(cm^2 * sr * um)) using a manufacturer-provided radiometric calibration file and dark reference data collected prior to each flight. Next, radiance hyperspectral images were radiometrically converted to units of reflectance using pixels within the white (~56% reflective) calibration tarp at each site. Specifically, Headwall's Spectral Angle Mapper classification tool was used to select 100 pixels within the tarp region and relate these pixels to a reference spectrum corresponding to the ~56% white tarp area. This reference spectrum is used to convert the rest of the pixels in each image to units of reflectance scaled between 0.0 to 1.0. The APX-15 trajectory for each flight was processed using Trimble R8s Base Station data (corrected using the National Oceanic and Atmospheric Administration's Online Positioning User Service) in Applanix POSPac UAV 8.8 software to yield a smoothed best estimate of trajectory (SBET) file to be used during orthorectification. In the Headwall SpectralView software, the reflectance hyperspectral images were orthorectified: this step removed image distortions introduced during flight and corrected for terrain displacement to assign geospatial coordinates to each hyperspectral image based on an input elevation and sensor trajectory data. We utilized a 1-meter digital elevation model (DEM) raster in GeoTIFF format downloaded from the USGS 3D Elevation Program (3DEP) lidar explorer (https://apps.nationalmap.gov/lidar-explorer/#/) along with each flight's respective SBET file. Using the Ortho-Rectification tool in SpectralView, parameters were iteratively tuned to obtain the best geometric corrections and alignment between multiple hyperspectral images. The orthorectified reflectance images are stored as 32-bit single precision floating point numbers in flat binary files with a band sequential (BSQ) interleave. Each image is accompanied by an ASCII text header file (.hdr) containing band center wavelengths and other parameters relevant to the images. The Headwall SpectralView software georeferences the hyperspectral images to a geographic coordinate system (latitude and longitude) and WGS84 datum with a spatial resolution of approximately 4 cm. To display a true color composite, use the following band combination: band 124 as red, band 65 as green, and band 38 as blue. 20250310 ECCOE Multispectral Imagery ELM Calibration Steps The Empirical Line Method (ELM; Smith and Milton, 1999) was used to calibrate the multispectral imagery processed using the manufacturer recommended methods (see Processing Step 3). Pure pixels from dark and bright calibration panels were manually selected from the NUSO processed multispectral MicaSense RedEdge-MX Dual imagery for each of the 10 spectral bands. The reflectance values measured by the field spectrometer for the dark and bright calibration panels were compared to the manually selected pure pixel values in the multispectral imagery to determine the relationship between them and produce updated calibration coefficients for each band. The updated coefficients were used to update the reflectance values for all pixels in the multispectral imagery. A .tif file was exported with the updated reflectance values for each of the 10 band multispectral bands. Python 3.10.9 was used to update the multispectral imagery. References Smith, G.M.; Milton, E.J. The use of the empirical line method to calibrate remotely sensed data to reflectance. International Journal of remote sensing 1999, 20, 2653-2662. 20250430 ECCOE Hyperspectral Imagery ELM Calibration Steps The Empirical Line Method (ELM; Smith and Milton, 1999) was used to calibrate the hyperspectral imagery processed using the manufacturer recommended methods (see Processing Step 4). Pure pixels from dark and bright calibration panels were manually selected from the NUSO processed hyperspectral Headwall Nano-Hyperspec intermediate step radiance imagery for each of the 274 spectral bands. The reflectance values measured by the field spectrometer for the dark and bright calibration panels were compared to the manually selected pure pixel values in the hyperspectral radiance imagery to determine the relationship between them and produce updated calibration coefficients for each band. The updated coefficients were used to update the radiance imagery to reflectance values for all pixels in the imagery. A .tif file was exported with the updated reflectance values for each of the 274 hyperspectral bands. Python 3.10.9 was used to update the hyperspectral imagery. References Smith, G.M.; Milton, E.J. The use of the empirical line method to calibrate remotely sensed data to reflectance. International Journal of remote sensing 1999, 20, 2653-2662. 20250430 Vector Point 0.00000167 0.00000167 Decimal degrees WGS84 WGS84 6378137.0 298.257223563 WGS84 0.1 meters Explicit elevation coordinate included with horizontal coordinates 2023-07-18_1200_ECCOE-EROS-GVR-Site_Single.csv Field collection data using the single spectroradiometer methodology. Please refer to the Data Dictionary for details on the formatting of the file. Producer Defined 2023-07-18_1200_ECCOE-EROS-GVR-Site_Dual.csv Field collection data using the dual spectroradiometer methodology. Please refer to the Data Dictionary for details on the formatting of the file. Producer Defined 2023-07-18_1200_ECCOE-EROS-GVR-Site_Single.gpkg Standardized GeoPackage database containing the field data in a vector format for the specified date and site. Please refer to the Data Dictionary for details on the formatting of the file. Producer Defined 2023-07-18_1200_ECCOE-EROS-GVR-Site_Dual.gpkg Standardized GeoPackage database containing the field data in a vector format for the specified date and site. Please refer to the Data Dictionary for details on the formatting of the file. Producer Defined 2023-07-18_1200_MicaSenseDual_ortho_panels.hdr Header file for associated TIF file, containing band names and center wavelength information for each of the 10 multispectral spectral bands. Producer Defined 2023-07-18_1200_MicaSenseDual_ortho_panels.tif 10 band raster geospatial data file. See data dictionary for further details. Producer Defined Value Unique numeric values contained in each raster cell. Producer Defined 0.0 0.8592529296875 2023-07-18_1200_MicaSenseDual_ortho_panels+sunSensor.hdr Header file for associated TIF file, containing band names and center wavelength information for each of the 10 multispectral spectral bands. Producer Defined 2023-07-18_1200_MicaSenseDual_ortho_panels+sunSensor.tif 10 band raster geospatial data file. See data dictionary for further details. Producer Defined Value Unique numeric values contained in each raster cell. Producer Defined 0.0 0.674652099609375 2023-07-18_1200_OrthoMosaic_Reflectance_HeadwallSpectrum.tif 274 band raster geospatial data file. See data dictionary for further details. Producer Defined Value Unique numeric values contained in each raster cell. Producer Defined 0.0 1.2254811525344849 2023-07-18_1200_MicaSenseDual_ortho_panels_Reflectance_ELM.tif 10 band raster geospatial data file. See data dictionary for further details. Producer Defined Value Unique numeric values contained in each raster cell. Producer Defined -0.012782808393239975 1.004025936126709 2023-07-18_1200_OrthoMosaic_Reflectance_Headwall_ELM.tif 274 band raster geospatial data file. See data dictionary for further details. Producer Defined Value Unique numeric values contained in each raster cell. Producer Defined -0.021690357476473 1.2730042934418 U.S. Geological Survey GS ScienceBase mailing address

Denver Federal Center, Building 810, Mail Stop 302

Denver CO 80225 United States 1-888-275-8747 sciencebase@usgs.gov Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data on any other system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty. Digital Data https://doi.org/10.5066/P14FFHW5 None 20251118 USGS EROS Center Customer Service Representative mailing and physical

47914 252nd St

Sioux Falls SD 57198-001 United States 1-800-252-4547 605-594-6589 custserv@usgs.gov FGDC Content Standard for Digital Geospatial Metadata FGDC-STD-001-1998